AutoML‐Multiverse: Model Agnostic Framework for Optimizing Individual‐level Alzheimer's Disease Classification

Preliminary findings reveal three key strengths of the AutoML-multiverse framework: (1) superior predictive performance over individual pipelines, (2) data-driven ensemble construction that reduces biases and arbitrary decisions, and (3) clinically relevant integration of neuroimaging, particularly for distinguishing slow from fast progressors, with potential utility in clinical trials. Future work will validate the robustness and generalisability of these results.